This paper considers the technical aspects and the consequences, in terms of simulation results and policy assessment, of introducing imperfect competition in a CGE model. The modifications to the standard CGE framework needed to model imperfect competition in some industries are briefly discussed. Next, the paper discusses whether, how much and why, those changes may affect the qualitative output of a typical simulation experiment. It is argued that technical choices made in designing the model structure may have a significant impact on the model behavior. This is especially evident when the output of the model, under an imperfect competition closure, is compared with that obtained under a standard closure, assuming perfect competition. As an illustration, a scenario of agricultural trade liberalization under alternative market structures is analyzed.